2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
13 * Based on ipv4/udp.c code.
15 * Authors: Vasiliy Kulikov / Openwall (for Linux 2.6),
16 * Pavel Kankovsky (for Linux 2.4.32)
18 * Pavel gave all rights to bugs to Vasiliy,
19 * none of the bugs are Pavel's now.
23 #include <linux/uaccess.h>
24 #include <linux/types.h>
25 #include <linux/fcntl.h>
26 #include <linux/socket.h>
27 #include <linux/sockios.h>
29 #include <linux/errno.h>
30 #include <linux/timer.h>
32 #include <linux/inet.h>
33 #include <linux/netdevice.h>
37 #include <net/protocol.h>
38 #include <linux/skbuff.h>
39 #include <linux/proc_fs.h>
40 #include <linux/export.h>
44 #include <net/route.h>
45 #include <net/inet_common.h>
46 #include <net/checksum.h>
48 #if IS_ENABLED(CONFIG_IPV6)
49 #include <linux/in6.h>
50 #include <linux/icmpv6.h>
51 #include <net/addrconf.h>
53 #include <net/transp_v6.h>
57 struct ping_table ping_table;
58 struct pingv6_ops pingv6_ops;
59 EXPORT_SYMBOL_GPL(pingv6_ops);
61 static u16 ping_port_rover;
63 static inline int ping_hashfn(struct net *net, unsigned int num, unsigned int mask)
65 int res = (num + net_hash_mix(net)) & mask;
67 pr_debug("hash(%d) = %d\n", num, res);
70 EXPORT_SYMBOL_GPL(ping_hash);
72 static inline struct hlist_nulls_head *ping_hashslot(struct ping_table *table,
73 struct net *net, unsigned int num)
75 return &table->hash[ping_hashfn(net, num, PING_HTABLE_MASK)];
78 int ping_get_port(struct sock *sk, unsigned short ident)
80 struct hlist_nulls_node *node;
81 struct hlist_nulls_head *hlist;
82 struct inet_sock *isk, *isk2;
83 struct sock *sk2 = NULL;
86 write_lock_bh(&ping_table.lock);
89 u16 result = ping_port_rover + 1;
91 for (i = 0; i < (1L << 16); i++, result++) {
93 result++; /* avoid zero */
94 hlist = ping_hashslot(&ping_table, sock_net(sk),
96 ping_portaddr_for_each_entry(sk2, node, hlist) {
99 if (isk2->inet_num == result)
104 ping_port_rover = ident = result;
112 hlist = ping_hashslot(&ping_table, sock_net(sk), ident);
113 ping_portaddr_for_each_entry(sk2, node, hlist) {
116 /* BUG? Why is this reuse and not reuseaddr? ping.c
117 * doesn't turn off SO_REUSEADDR, and it doesn't expect
118 * that other ping processes can steal its packets.
120 if ((isk2->inet_num == ident) &&
122 (!sk2->sk_reuse || !sk->sk_reuse))
127 pr_debug("found port/ident = %d\n", ident);
128 isk->inet_num = ident;
129 if (sk_unhashed(sk)) {
130 pr_debug("was not hashed\n");
132 hlist_nulls_add_head(&sk->sk_nulls_node, hlist);
133 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
135 write_unlock_bh(&ping_table.lock);
139 write_unlock_bh(&ping_table.lock);
142 EXPORT_SYMBOL_GPL(ping_get_port);
144 void ping_hash(struct sock *sk)
146 pr_debug("ping_hash(sk->port=%u)\n", inet_sk(sk)->inet_num);
147 BUG(); /* "Please do not press this button again." */
150 void ping_unhash(struct sock *sk)
152 struct inet_sock *isk = inet_sk(sk);
153 pr_debug("ping_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
155 write_lock_bh(&ping_table.lock);
156 hlist_nulls_del(&sk->sk_nulls_node);
160 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
161 write_unlock_bh(&ping_table.lock);
164 EXPORT_SYMBOL_GPL(ping_unhash);
166 static struct sock *ping_lookup(struct net *net, struct sk_buff *skb, u16 ident)
168 struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
169 struct sock *sk = NULL;
170 struct inet_sock *isk;
171 struct hlist_nulls_node *hnode;
172 int dif = skb->dev->ifindex;
174 if (skb->protocol == htons(ETH_P_IP)) {
175 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
176 (int)ident, &ip_hdr(skb)->daddr, dif);
177 #if IS_ENABLED(CONFIG_IPV6)
178 } else if (skb->protocol == htons(ETH_P_IPV6)) {
179 pr_debug("try to find: num = %d, daddr = %pI6c, dif = %d\n",
180 (int)ident, &ipv6_hdr(skb)->daddr, dif);
184 read_lock_bh(&ping_table.lock);
186 ping_portaddr_for_each_entry(sk, hnode, hslot) {
189 pr_debug("iterate\n");
190 if (isk->inet_num != ident)
193 if (skb->protocol == htons(ETH_P_IP) &&
194 sk->sk_family == AF_INET) {
195 pr_debug("found: %p: num=%d, daddr=%pI4, dif=%d\n", sk,
196 (int) isk->inet_num, &isk->inet_rcv_saddr,
197 sk->sk_bound_dev_if);
199 if (isk->inet_rcv_saddr &&
200 isk->inet_rcv_saddr != ip_hdr(skb)->daddr)
202 #if IS_ENABLED(CONFIG_IPV6)
203 } else if (skb->protocol == htons(ETH_P_IPV6) &&
204 sk->sk_family == AF_INET6) {
205 struct ipv6_pinfo *np = inet6_sk(sk);
207 pr_debug("found: %p: num=%d, daddr=%pI6c, dif=%d\n", sk,
209 &inet6_sk(sk)->rcv_saddr,
210 sk->sk_bound_dev_if);
212 if (!ipv6_addr_any(&np->rcv_saddr) &&
213 !ipv6_addr_equal(&np->rcv_saddr,
214 &ipv6_hdr(skb)->daddr))
219 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
228 read_unlock_bh(&ping_table.lock);
233 static void inet_get_ping_group_range_net(struct net *net, kgid_t *low,
236 kgid_t *data = net->ipv4.sysctl_ping_group_range;
240 seq = read_seqbegin(&sysctl_local_ports.lock);
244 } while (read_seqretry(&sysctl_local_ports.lock, seq));
248 int ping_init_sock(struct sock *sk)
250 struct net *net = sock_net(sk);
251 kgid_t group = current_egid();
252 struct group_info *group_info;
257 inet_get_ping_group_range_net(net, &low, &high);
258 if (gid_lte(low, group) && gid_lte(group, high))
261 group_info = get_current_groups();
262 count = group_info->ngroups;
263 for (i = 0; i < group_info->nblocks; i++) {
264 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
265 for (j = 0; j < cp_count; j++) {
266 kgid_t gid = group_info->blocks[i][j];
267 if (gid_lte(low, gid) && gid_lte(gid, high))
268 goto out_release_group;
277 put_group_info(group_info);
280 EXPORT_SYMBOL_GPL(ping_init_sock);
282 void ping_close(struct sock *sk, long timeout)
284 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
285 inet_sk(sk), inet_sk(sk)->inet_num);
286 pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
288 sk_common_release(sk);
290 EXPORT_SYMBOL_GPL(ping_close);
292 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
293 int ping_check_bind_addr(struct sock *sk, struct inet_sock *isk,
294 struct sockaddr *uaddr, int addr_len) {
295 struct net *net = sock_net(sk);
296 if (sk->sk_family == AF_INET) {
297 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
300 if (addr_len < sizeof(*addr))
303 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI4,port=%d)\n",
304 sk, &addr->sin_addr.s_addr, ntohs(addr->sin_port));
306 chk_addr_ret = inet_addr_type(net, addr->sin_addr.s_addr);
308 if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
309 chk_addr_ret = RTN_LOCAL;
311 if ((sysctl_ip_nonlocal_bind == 0 &&
312 isk->freebind == 0 && isk->transparent == 0 &&
313 chk_addr_ret != RTN_LOCAL) ||
314 chk_addr_ret == RTN_MULTICAST ||
315 chk_addr_ret == RTN_BROADCAST)
316 return -EADDRNOTAVAIL;
318 #if IS_ENABLED(CONFIG_IPV6)
319 } else if (sk->sk_family == AF_INET6) {
320 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) uaddr;
321 int addr_type, scoped, has_addr;
322 struct net_device *dev = NULL;
324 if (addr_len < sizeof(*addr))
327 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI6c,port=%d)\n",
328 sk, addr->sin6_addr.s6_addr, ntohs(addr->sin6_port));
330 addr_type = ipv6_addr_type(&addr->sin6_addr);
331 scoped = __ipv6_addr_needs_scope_id(addr_type);
332 if ((addr_type != IPV6_ADDR_ANY &&
333 !(addr_type & IPV6_ADDR_UNICAST)) ||
334 (scoped && !addr->sin6_scope_id))
338 if (addr->sin6_scope_id) {
339 dev = dev_get_by_index_rcu(net, addr->sin6_scope_id);
345 has_addr = pingv6_ops.ipv6_chk_addr(net, &addr->sin6_addr, dev,
349 if (!(isk->freebind || isk->transparent || has_addr ||
350 addr_type == IPV6_ADDR_ANY))
351 return -EADDRNOTAVAIL;
354 sk->sk_bound_dev_if = addr->sin6_scope_id;
357 return -EAFNOSUPPORT;
362 void ping_set_saddr(struct sock *sk, struct sockaddr *saddr)
364 if (saddr->sa_family == AF_INET) {
365 struct inet_sock *isk = inet_sk(sk);
366 struct sockaddr_in *addr = (struct sockaddr_in *) saddr;
367 isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
368 #if IS_ENABLED(CONFIG_IPV6)
369 } else if (saddr->sa_family == AF_INET6) {
370 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) saddr;
371 struct ipv6_pinfo *np = inet6_sk(sk);
372 np->rcv_saddr = np->saddr = addr->sin6_addr;
377 void ping_clear_saddr(struct sock *sk, int dif)
379 sk->sk_bound_dev_if = dif;
380 if (sk->sk_family == AF_INET) {
381 struct inet_sock *isk = inet_sk(sk);
382 isk->inet_rcv_saddr = isk->inet_saddr = 0;
383 #if IS_ENABLED(CONFIG_IPV6)
384 } else if (sk->sk_family == AF_INET6) {
385 struct ipv6_pinfo *np = inet6_sk(sk);
386 memset(&np->rcv_saddr, 0, sizeof(np->rcv_saddr));
387 memset(&np->saddr, 0, sizeof(np->saddr));
392 * We need our own bind because there are no privileged id's == local ports.
393 * Moreover, we don't allow binding to multi- and broadcast addresses.
396 int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
398 struct inet_sock *isk = inet_sk(sk);
401 int dif = sk->sk_bound_dev_if;
403 err = ping_check_bind_addr(sk, isk, uaddr, addr_len);
410 if (isk->inet_num != 0)
414 ping_set_saddr(sk, uaddr);
415 snum = ntohs(((struct sockaddr_in *)uaddr)->sin_port);
416 if (ping_get_port(sk, snum) != 0) {
417 ping_clear_saddr(sk, dif);
421 pr_debug("after bind(): num = %d, dif = %d\n",
423 (int)sk->sk_bound_dev_if);
426 if ((sk->sk_family == AF_INET && isk->inet_rcv_saddr) ||
427 (sk->sk_family == AF_INET6 &&
428 !ipv6_addr_any(&inet6_sk(sk)->rcv_saddr)))
429 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
432 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
433 isk->inet_sport = htons(isk->inet_num);
437 #if IS_ENABLED(CONFIG_IPV6)
438 if (sk->sk_family == AF_INET6)
439 memset(&inet6_sk(sk)->daddr, 0, sizeof(inet6_sk(sk)->daddr));
445 pr_debug("ping_v4_bind -> %d\n", err);
448 EXPORT_SYMBOL_GPL(ping_bind);
451 * Is this a supported type of ICMP message?
454 static inline int ping_supported(int family, int type, int code)
456 return (family == AF_INET && type == ICMP_ECHO && code == 0) ||
457 (family == AF_INET6 && type == ICMPV6_ECHO_REQUEST && code == 0);
461 * This routine is called by the ICMP module when it gets some
462 * sort of error condition.
465 void ping_err(struct sk_buff *skb, int offset, u32 info)
468 struct icmphdr *icmph;
469 struct inet_sock *inet_sock;
472 struct net *net = dev_net(skb->dev);
477 if (skb->protocol == htons(ETH_P_IP)) {
479 type = icmp_hdr(skb)->type;
480 code = icmp_hdr(skb)->code;
481 icmph = (struct icmphdr *)(skb->data + offset);
482 } else if (skb->protocol == htons(ETH_P_IPV6)) {
484 type = icmp6_hdr(skb)->icmp6_type;
485 code = icmp6_hdr(skb)->icmp6_code;
486 icmph = (struct icmphdr *) (skb->data + offset);
491 /* We assume the packet has already been checked by icmp_unreach */
493 if (!ping_supported(family, icmph->type, icmph->code))
496 pr_debug("ping_err(proto=0x%x,type=%d,code=%d,id=%04x,seq=%04x)\n",
497 skb->protocol, type, code, ntohs(icmph->un.echo.id),
498 ntohs(icmph->un.echo.sequence));
500 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
502 pr_debug("no socket, dropping\n");
503 return; /* No socket for error */
505 pr_debug("err on socket %p\n", sk);
509 inet_sock = inet_sk(sk);
511 if (skb->protocol == htons(ETH_P_IP)) {
514 case ICMP_TIME_EXCEEDED:
517 case ICMP_SOURCE_QUENCH:
518 /* This is not a real error but ping wants to see it.
519 * Report it with some fake errno.
523 case ICMP_PARAMETERPROB:
527 case ICMP_DEST_UNREACH:
528 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
529 ipv4_sk_update_pmtu(skb, sk, info);
530 if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
538 if (code <= NR_ICMP_UNREACH) {
539 harderr = icmp_err_convert[code].fatal;
540 err = icmp_err_convert[code].errno;
544 /* See ICMP_SOURCE_QUENCH */
545 ipv4_sk_redirect(skb, sk);
549 #if IS_ENABLED(CONFIG_IPV6)
550 } else if (skb->protocol == htons(ETH_P_IPV6)) {
551 harderr = pingv6_ops.icmpv6_err_convert(type, code, &err);
556 * RFC1122: OK. Passes ICMP errors back to application, as per
559 if ((family == AF_INET && !inet_sock->recverr) ||
560 (family == AF_INET6 && !inet6_sk(sk)->recverr)) {
561 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
564 if (family == AF_INET) {
565 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
567 #if IS_ENABLED(CONFIG_IPV6)
568 } else if (family == AF_INET6) {
569 pingv6_ops.ipv6_icmp_error(sk, skb, err, 0,
575 sk->sk_error_report(sk);
579 EXPORT_SYMBOL_GPL(ping_err);
582 * Copy and checksum an ICMP Echo packet from user space into a buffer
583 * starting from the payload.
586 int ping_getfrag(void *from, char *to,
587 int offset, int fraglen, int odd, struct sk_buff *skb)
589 struct pingfakehdr *pfh = (struct pingfakehdr *)from;
592 if (fraglen < sizeof(struct icmphdr))
594 if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
595 pfh->iov, 0, fraglen - sizeof(struct icmphdr),
598 } else if (offset < sizeof(struct icmphdr)) {
601 if (csum_partial_copy_fromiovecend
602 (to, pfh->iov, offset - sizeof(struct icmphdr),
603 fraglen, &pfh->wcheck))
607 #if IS_ENABLED(CONFIG_IPV6)
608 /* For IPv6, checksum each skb as we go along, as expected by
609 * icmpv6_push_pending_frames. For IPv4, accumulate the checksum in
610 * wcheck, it will be finalized in ping_v4_push_pending_frames.
612 if (pfh->family == AF_INET6) {
613 skb->csum = pfh->wcheck;
614 skb->ip_summed = CHECKSUM_NONE;
621 EXPORT_SYMBOL_GPL(ping_getfrag);
623 static int ping_v4_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
626 struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
628 pfh->wcheck = csum_partial((char *)&pfh->icmph,
629 sizeof(struct icmphdr), pfh->wcheck);
630 pfh->icmph.checksum = csum_fold(pfh->wcheck);
631 memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
632 skb->ip_summed = CHECKSUM_NONE;
633 return ip_push_pending_frames(sk, fl4);
636 int ping_common_sendmsg(int family, struct msghdr *msg, size_t len,
637 void *user_icmph, size_t icmph_len) {
647 /* Mirror BSD error message compatibility */
648 if (msg->msg_flags & MSG_OOB)
652 * Fetch the ICMP header provided by the userland.
653 * iovec is modified! The ICMP header is consumed.
655 if (memcpy_fromiovec(user_icmph, msg->msg_iov, icmph_len))
658 if (family == AF_INET) {
659 type = ((struct icmphdr *) user_icmph)->type;
660 code = ((struct icmphdr *) user_icmph)->code;
661 #if IS_ENABLED(CONFIG_IPV6)
662 } else if (family == AF_INET6) {
663 type = ((struct icmp6hdr *) user_icmph)->icmp6_type;
664 code = ((struct icmp6hdr *) user_icmph)->icmp6_code;
670 if (!ping_supported(family, type, code))
675 EXPORT_SYMBOL_GPL(ping_common_sendmsg);
677 int ping_v4_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
680 struct net *net = sock_net(sk);
682 struct inet_sock *inet = inet_sk(sk);
683 struct ipcm_cookie ipc;
684 struct icmphdr user_icmph;
685 struct pingfakehdr pfh;
686 struct rtable *rt = NULL;
687 struct ip_options_data opt_copy;
689 __be32 saddr, daddr, faddr;
693 pr_debug("ping_v4_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
695 err = ping_common_sendmsg(AF_INET, msg, len, &user_icmph,
701 * Get and verify the address.
705 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
706 if (msg->msg_namelen < sizeof(*usin))
708 if (usin->sin_family != AF_INET)
710 daddr = usin->sin_addr.s_addr;
713 if (sk->sk_state != TCP_ESTABLISHED)
714 return -EDESTADDRREQ;
715 daddr = inet->inet_daddr;
719 ipc.addr = inet->inet_saddr;
721 ipc.oif = sk->sk_bound_dev_if;
724 sock_tx_timestamp(sk, &ipc.tx_flags);
726 if (msg->msg_controllen) {
727 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
734 struct ip_options_rcu *inet_opt;
737 inet_opt = rcu_dereference(inet->inet_opt);
739 memcpy(&opt_copy, inet_opt,
740 sizeof(*inet_opt) + inet_opt->opt.optlen);
741 ipc.opt = &opt_copy.opt;
747 ipc.addr = faddr = daddr;
749 if (ipc.opt && ipc.opt->opt.srr) {
752 faddr = ipc.opt->opt.faddr;
754 tos = RT_TOS(inet->tos);
755 if (sock_flag(sk, SOCK_LOCALROUTE) ||
756 (msg->msg_flags & MSG_DONTROUTE) ||
757 (ipc.opt && ipc.opt->opt.is_strictroute)) {
761 if (ipv4_is_multicast(daddr)) {
763 ipc.oif = inet->mc_index;
765 saddr = inet->mc_addr;
767 ipc.oif = inet->uc_index;
769 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
770 RT_SCOPE_UNIVERSE, sk->sk_protocol,
771 inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
773 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
774 rt = ip_route_output_flow(net, &fl4, sk);
778 if (err == -ENETUNREACH)
779 IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
784 if ((rt->rt_flags & RTCF_BROADCAST) &&
785 !sock_flag(sk, SOCK_BROADCAST))
788 if (msg->msg_flags & MSG_CONFIRM)
793 ipc.addr = fl4.daddr;
797 pfh.icmph.type = user_icmph.type; /* already checked */
798 pfh.icmph.code = user_icmph.code; /* ditto */
799 pfh.icmph.checksum = 0;
800 pfh.icmph.un.echo.id = inet->inet_sport;
801 pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
802 pfh.iov = msg->msg_iov;
804 pfh.family = AF_INET;
806 err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
807 0, &ipc, &rt, msg->msg_flags);
809 ip_flush_pending_frames(sk);
811 err = ping_v4_push_pending_frames(sk, &pfh, &fl4);
819 icmp_out_count(sock_net(sk), user_icmph.type);
825 dst_confirm(&rt->dst);
826 if (!(msg->msg_flags & MSG_PROBE) || len)
827 goto back_from_confirm;
832 int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
833 size_t len, int noblock, int flags, int *addr_len)
835 struct inet_sock *isk = inet_sk(sk);
836 int family = sk->sk_family;
837 struct sockaddr_in *sin;
838 struct sockaddr_in6 *sin6;
842 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
849 if (family == AF_INET)
850 *addr_len = sizeof(*sin);
851 else if (family == AF_INET6 && addr_len)
852 *addr_len = sizeof(*sin6);
855 if (flags & MSG_ERRQUEUE) {
856 if (family == AF_INET) {
857 return ip_recv_error(sk, msg, len);
858 #if IS_ENABLED(CONFIG_IPV6)
859 } else if (family == AF_INET6) {
860 return pingv6_ops.ipv6_recv_error(sk, msg, len);
865 skb = skb_recv_datagram(sk, flags, noblock, &err);
871 msg->msg_flags |= MSG_TRUNC;
875 /* Don't bother checking the checksum */
876 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
880 sock_recv_timestamp(msg, sk, skb);
882 /* Copy the address and add cmsg data. */
883 if (family == AF_INET) {
884 sin = (struct sockaddr_in *) msg->msg_name;
886 sin->sin_family = AF_INET;
887 sin->sin_port = 0 /* skb->h.uh->source */;
888 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
889 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
893 ip_cmsg_recv(msg, skb);
895 #if IS_ENABLED(CONFIG_IPV6)
896 } else if (family == AF_INET6) {
897 struct ipv6_pinfo *np = inet6_sk(sk);
898 struct ipv6hdr *ip6 = ipv6_hdr(skb);
899 sin6 = (struct sockaddr_in6 *) msg->msg_name;
902 sin6->sin6_family = AF_INET6;
904 sin6->sin6_addr = ip6->saddr;
905 sin6->sin6_flowinfo = 0;
907 sin6->sin6_flowinfo = ip6_flowinfo(ip6);
908 sin6->sin6_scope_id =
909 ipv6_iface_scope_id(&sin6->sin6_addr,
913 if (inet6_sk(sk)->rxopt.all)
914 pingv6_ops.ip6_datagram_recv_ctl(sk, msg, skb);
923 skb_free_datagram(sk, skb);
925 pr_debug("ping_recvmsg -> %d\n", err);
928 EXPORT_SYMBOL_GPL(ping_recvmsg);
930 int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
932 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
933 inet_sk(sk), inet_sk(sk)->inet_num, skb);
934 if (sock_queue_rcv_skb(sk, skb) < 0) {
936 pr_debug("ping_queue_rcv_skb -> failed\n");
941 EXPORT_SYMBOL_GPL(ping_queue_rcv_skb);
945 * All we need to do is get the socket.
948 void ping_rcv(struct sk_buff *skb)
951 struct net *net = dev_net(skb->dev);
952 struct icmphdr *icmph = icmp_hdr(skb);
954 /* We assume the packet has already been checked by icmp_rcv */
956 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
957 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
959 /* Push ICMP header back */
960 skb_push(skb, skb->data - (u8 *)icmph);
962 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
964 pr_debug("rcv on socket %p\n", sk);
965 ping_queue_rcv_skb(sk, skb_get(skb));
969 pr_debug("no socket, dropping\n");
971 /* We're called from icmp_rcv(). kfree_skb() is done there. */
973 EXPORT_SYMBOL_GPL(ping_rcv);
975 struct proto ping_prot = {
977 .owner = THIS_MODULE,
978 .init = ping_init_sock,
980 .connect = ip4_datagram_connect,
981 .disconnect = udp_disconnect,
982 .setsockopt = ip_setsockopt,
983 .getsockopt = ip_getsockopt,
984 .sendmsg = ping_v4_sendmsg,
985 .recvmsg = ping_recvmsg,
987 .backlog_rcv = ping_queue_rcv_skb,
988 .release_cb = ip4_datagram_release_cb,
990 .unhash = ping_unhash,
991 .get_port = ping_get_port,
992 .obj_size = sizeof(struct inet_sock),
994 EXPORT_SYMBOL(ping_prot);
996 #ifdef CONFIG_PROC_FS
998 static struct sock *ping_get_first(struct seq_file *seq, int start)
1001 struct ping_iter_state *state = seq->private;
1002 struct net *net = seq_file_net(seq);
1004 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
1006 struct hlist_nulls_node *node;
1007 struct hlist_nulls_head *hslot;
1009 hslot = &ping_table.hash[state->bucket];
1011 if (hlist_nulls_empty(hslot))
1014 sk_nulls_for_each(sk, node, hslot) {
1015 if (net_eq(sock_net(sk), net))
1024 static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
1026 struct ping_iter_state *state = seq->private;
1027 struct net *net = seq_file_net(seq);
1030 sk = sk_nulls_next(sk);
1031 } while (sk && (!net_eq(sock_net(sk), net)));
1034 return ping_get_first(seq, state->bucket + 1);
1038 static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
1040 struct sock *sk = ping_get_first(seq, 0);
1043 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
1045 return pos ? NULL : sk;
1048 static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
1050 struct ping_iter_state *state = seq->private;
1053 read_lock_bh(&ping_table.lock);
1055 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
1058 static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1062 if (v == SEQ_START_TOKEN)
1063 sk = ping_get_idx(seq, 0);
1065 sk = ping_get_next(seq, v);
1071 static void ping_seq_stop(struct seq_file *seq, void *v)
1073 read_unlock_bh(&ping_table.lock);
1076 static void ping_format_sock(struct sock *sp, struct seq_file *f,
1077 int bucket, int *len)
1079 struct inet_sock *inet = inet_sk(sp);
1080 __be32 dest = inet->inet_daddr;
1081 __be32 src = inet->inet_rcv_saddr;
1082 __u16 destp = ntohs(inet->inet_dport);
1083 __u16 srcp = ntohs(inet->inet_sport);
1085 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
1086 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
1087 bucket, src, srcp, dest, destp, sp->sk_state,
1088 sk_wmem_alloc_get(sp),
1089 sk_rmem_alloc_get(sp),
1091 from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
1093 atomic_read(&sp->sk_refcnt), sp,
1094 atomic_read(&sp->sk_drops), len);
1097 static int ping_seq_show(struct seq_file *seq, void *v)
1099 if (v == SEQ_START_TOKEN)
1100 seq_printf(seq, "%-127s\n",
1101 " sl local_address rem_address st tx_queue "
1102 "rx_queue tr tm->when retrnsmt uid timeout "
1103 "inode ref pointer drops");
1105 struct ping_iter_state *state = seq->private;
1108 ping_format_sock(v, seq, state->bucket, &len);
1109 seq_printf(seq, "%*s\n", 127 - len, "");
1114 static const struct seq_operations ping_seq_ops = {
1115 .show = ping_seq_show,
1116 .start = ping_seq_start,
1117 .next = ping_seq_next,
1118 .stop = ping_seq_stop,
1121 static int ping_seq_open(struct inode *inode, struct file *file)
1123 return seq_open_net(inode, file, &ping_seq_ops,
1124 sizeof(struct ping_iter_state));
1127 static const struct file_operations ping_seq_fops = {
1128 .open = ping_seq_open,
1130 .llseek = seq_lseek,
1131 .release = seq_release_net,
1134 static int ping_proc_register(struct net *net)
1136 struct proc_dir_entry *p;
1139 p = proc_create("icmp", S_IRUGO, net->proc_net, &ping_seq_fops);
1145 static void ping_proc_unregister(struct net *net)
1147 remove_proc_entry("icmp", net->proc_net);
1151 static int __net_init ping_proc_init_net(struct net *net)
1153 return ping_proc_register(net);
1156 static void __net_exit ping_proc_exit_net(struct net *net)
1158 ping_proc_unregister(net);
1161 static struct pernet_operations ping_net_ops = {
1162 .init = ping_proc_init_net,
1163 .exit = ping_proc_exit_net,
1166 int __init ping_proc_init(void)
1168 return register_pernet_subsys(&ping_net_ops);
1171 void ping_proc_exit(void)
1173 unregister_pernet_subsys(&ping_net_ops);
1178 void __init ping_init(void)
1182 for (i = 0; i < PING_HTABLE_SIZE; i++)
1183 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
1184 rwlock_init(&ping_table.lock);